Defense AI is not a topic that is overtly discussed in Spain. Instead of reflecting on the usage of defense AI, governmental speeches often acknowledge that defense AI is no longer a separate issue. Rather, defense AI is an element that is part of a broader range of issues revolving around a holistic approach to national security.

An analytical assessment of existing national strategies, technology-oriented industrial defense plans, and directives on defense policy show that until around 2020, AI was not a central element in the Spanish defense ecosystem. Until then, the general importance of AI had been acknowledged, while risks and benefits had been generally addressed. However, it was not until the start of the new decade of the 2020s that AI took center stage with specific principles, roadmaps, and several initiatives outlining how AI should be governed and implemented. Public documents also acknowledge that the Spanish ecosystem has lagged in the development of defense AI. This, the documents contend, has negatively affected Spain’s international positioning on AI and limited the competitiveness of the local defense industry—something that ought to be addressed.

Spain’s governance system for defense AI includes mandates and processes to coordinate stakeholders across relevant institutions. An increase in funding, a diversification of projects, and access to much more publicly available documents on the descriptive plans of projects as well as their final results and impact may be convenient to create a culture of defense AI and are potential policy recommendations to move forward.

A small number of large companies lead the defense AI market and act as leaders of several relevant European Union (EU) projects. Still, this article proposes that a larger focus should be put onto small and medium-sized enterprises (SMEs) to ensure that defense AI emerges from a diversified portfolio of services across different types of firms. This would help increase the defense industry market footprint in Spain, expand and deepen existing value chains while also creating new ones across the industry, and increase the level of involvement of Spanish companies in European projects.

Defense AI is fielded and operationalized via different instruments and takes place in many strategic sectors that Spain has flagged for national security (such as aerospace, cybersecurity, and maritime security). However, further activities are needed to fully leverage synergies with other technologies (e.g. quantum) in research and development (R&D) projects.

Finally, there is limited information on the training of systems and human capital. Based on existing national and European regulation as well as the respect for fundamental rights, university programs should more actively include defense AI in their curricula.

1 Thinking About Defense AI

Defense AI is not a topic that is overtly discussed in Spain. Instead of reflecting on the usage of defense AI, governmental speeches often acknowledge that defense AI is no longer a separate issue. Rather, defense AI is an element that is part of a broader range of issues revolving around a holistic approach to national security.

There is no general definition of the meaning and the potential use cases of AI in support of Spain’s national security and defense. The lack, however, has not created problems but rather stems from the fact that Spain had been waiting for an EU-wide definition of AI provided by the AI Act adopted under the Spanish Presidency to the Council of the European Union in the second half of 2023. Spain supports a human-centric approach to AI in general, especially a “no harm” policy, respect for personal data privacy and support for human rights. However, Spain has, thus far, hardly addressed the ethics of defense AI as the dominating AI ethics debate excludes defense. Nevertheless, the fact that the whole-of-government approach to AI is human-centric suggests that any defense AI development will need to rely on ethical principles that govern AI in general.

An analytical assessment of existing national strategies, technology-oriented industrial defense plans and directives on defense policy show that, until around 2020, AI did not play a prominent role in Spain’s defense discourse. While AI was always acknowledged as an important technology, and most capstone documents underlined the need to address its challenges and opportunities, it was not until the start of the new decade that AI was listed as a central element, and given a set of measures, grounded principles, roadmaps, actions or at least considerations on how it should be governed. In general, Spain has followed a human-centric approach to AI with a strong emphasis on regulation, while guiding the work on the AI Act proposal, but this line of reasoning has been completely separated from defense concerns. Rather, Spain’s defense community has developed its own thinking on defense AI via the respective channels in the EU and NATO.

Public documents also acknowledge that the Spanish ecosystem has lagged in the development of defense AI. They also explicitly state that this situation has limited Spain’s positioning globally and the development of an economically competitive AI-capable defense industry.

1.1 A Broader Outlook on the National Security Strategy

The role and importance of AI in the defense and national security ecosystem in Spain has evolved in recent years. In the National Security Strategy from 2017 (National Security Department 2017), AI was mentioned only twice and in broad statements that reflected the importance of this topic from an intellectual perspective. No specific policy action, roadmap or initiative was drawn. However, with the renewal of the National Security Strategy (NSS) in 2021 (National Security Department 2021), AI turned from being an intellectual topic into becoming a central element of the ways, means and goals, as well as instruments, that frame, explain and provide guidance to the NSS.

Technology and strategic sectors are ranked second out of three main goals of the 2021 NSS. The strategy broadens (Ricart 2022) the scope and identifies technology, not only as a driver to boost the security and defense dimension, but also an end in itself, which needs to be underpinned with adequate industrial capabilities. The strategy emphasizes the key role of technology but has also been criticized as being too generic and lacking substance on the goals to be achieved and the roadmaps to implement them (Arteaga 2022a, b).

Particularly, AI is perceived as the most strategic technology across all sectors. The 2021 NSS devotes great attention to this technology and approaches it in two ways: as a power tool, and as an enabling, facilitating tool.

As a power tool, AI has become a key geoeconomic challenge. Spain explicitly mentions that China, which has achieved a mature development level in AI, aims to gain a predominant position that allows the country to define international standards and technical and industrial protocols. It wants to become the powerhouse leading foreign direct investments and technology flows into other countries to gain influence and market share in services and network operators. Second, AI is also a power tool because most developments of digital services and infrastructures rely on emerging and disruptive technologies such as cloud computing, quantum technologies, network visualization and the Internet of Things, all of which largely depend on the maturity of AI. Third, AI poses risks that are interrelated. Threats derived from the usage of next-generation technologies, such as AI, add complexity and impinge on individual rights through the illicit usage of AI. Finally, the 2021 NSS makes an explicit reference to the respect for human rights and ethics when designing, developing, deploying, and using AI, such as in the case of lethal autonomous weapons systems (LAWS). Although there is no real public debate about the ethics of LAWS, strategies always refer to the need to support arms control for AI and limit harm while not curtailing necessary military innovation.

Second, as an enabling and facilitating tool, AI is perceived as an opportunity that may be leveraged for specific applications. While the strategy acknowledges that AI may apply to a large range of case uses, it highlights two specific policy areas: counterintelligence, the fight against disinformation campaigns and actions against foreign interference; and maritime security. Regarding counterintelligence, it calls for the human and technological capacities to be strengthened, to continue to reap the benefits linked to proper data management and processing, such as AI, quantum computing and the cloud. In addition, the strategy calls for an update to legislation to guarantee the data protection rights of Spanish citizens, while enhancing the capacity of the Intelligence Services to act in their defense. On maritime security, the strategy sets the safety of the Spanish merchant and fishing fleet in Spanish jurisdictional and international waters as a priority. In this regard predictive analytics supports effective planning for complex operational scenarios. That’s why AI plays an increasingly important role in maritime surveillance systems, platforms, and sensors to modernize maritime capabilities.

AI is not only seen as a major element of national security and defense. AI also supports the three objectives of the 2021 NSS:

  • Advancing crisis management by adopting an anticipatory approach and supporting data-driven decision-making, across the central government and in co-governance with all regional authorities.

  • Strengthening technological capabilities and strategic sectors by adapting and updating strategic sectors to the much-needed quality controls, regulatory adaptation and research and developments (R&D) projects that may be accelerated and facilitated by AI.

  • Developing capabilities to prevent, deter, and respond to hybrid threats, where AI may be an active part of malicious hybrid threats, but also an active part in the fight against them.

In a nutshell, the three Ps of the 2021 National Security Strategy are to Protect, Promote and Participate—with AI playing a foundational role across all three elements. To achieve the 3 Ps, the security ecosystem would need to integrate AI across various instruments. However, AI is not explicitly mentioned in the strategy with regards to the other remaining instruments: resources on national security, preparation plans, indicator-based early warning systems, development of special communications from the Government Presidency, and the integration of regional authorities (“Comunidades y Ciudades Autónomas”) into the national security system.

1.2 The Directives on National Defense and on Defense Policy

While the national security strategy emphasizes AI as an important topic, the Directive on National Defense (DND) (Ministry of Defense 2020a) does so only limitedly. The DND is Spain’s top defense planning document and occupies the main position in Spanish defense policy after the Organic Law on National Defense 5/2005. The common purpose of the DND is to establish the objectives and the main lines of effort pursued by the Ministry of Defense (MoD). Among the fourteen goals mentioned in the document, technology ranks twelfth—behind issues like climate change and gender equality. Descriptions remain vague, encapsulating no specific policy actions or initiatives.

Despite this low position in the list of objectives, the DND highlights the need to foster R&D and innovation in technological capabilities. Also, it stresses the importance to foster capacity to attract technology-skilled talent i.e. through training, education, and skilled immigration policies. While capacity and specialization are needed, the DND highlights the importance of having “availability” of AI to be used, always based on national legislation. However, statements on AI are very limited, and there is no prominent importance given to AI in comparison to other technologies.

In the Guidelines, section 7 speaks of promoting the EU’s Common Foreign and Security Policy (CFSP), including industrial and technological cooperation, but only superficially. The second to last guideline (14th) is reserved for promoting the Industrial and Technological Base.

To give a further understanding of the practical implications of the DND, it is important to assess the Directive on Defense Policy (DDP) (Ministry of Defense 2020b), which develops and implements the guidelines derived from the DND in the MoD’s area of responsibility. It was published for the first time in 2020 with the goal of providing details on the implementation of the DND, which had already existed since 2010 but without a publicly available roadmap. The DDP fills this void.

Although the DND does not identify technology as a major topic, the DDP considers technology as a cross-cutting phenomenon that needs to be addressed in three ways: through a trust-based approach, by empowering strategic autonomy and reducing dependence in critical value chains, and by fostering a more cooperative European defense industrial and technological base (DITB). Against this background, the DDP’s principles fall into two categories. First, the DDP outlines the strengthening of the Spanish DTIB by

  1. 1.

    facilitating the development of dual-use technologies with other ministries, particularly the Ministry of Industry and Tourism, the Ministry of Science, Innovation and Universities, and the Ministry of Treasury and Public Administration, as long as operational requirements allow it;

  2. 2.

    prioritizing the use of technology and training over the large-scale deployment of human resources;

  3. 3.

    accomplishing greater strategic performance with the least personnel and material effort;

  4. 4.

    seeking maximum flexibility in dealing with multi-purpose and multi-domain tasks;

  5. 5.

    advancing the development of capabilities aimed at neutralizing hybrid threats.

Second, it aims to strengthen the internationalization of the national defense industry by

  1. 1.

    promoting industrial and technological cooperation as one of the main elements of Defense Diplomacy;

  2. 2.

    identifying the EU and NATO as priority frameworks for the internationalization of Spain’s defense industry;

  3. 3.

    making an integrated and coordinated effort, seeking the collaboration of all ministerial departments.

However, to implement specific measures for defense AI, the Directive acknowledges that the most important asset will be to deploy “reasonable financial resources” and maintain a stable budget. Defense investment should be approached in such a way that it not only meets the needs of public service, but also contributes to innovation, technological development, job creation and the projection of Spain’s influence in the world.

1.3 The Specific Focus of the 2020 Strategy on Technology and Innovation in Defense (ETID 2020)

The ETID (Ministry of Defense 2020c) was positively received. It did not produce a novel strategy but updated the one from 2015. The update reflected the need to explore upon new technologies and their impact on defense capabilities. Overall, the document rests on three pillars of

  • directing research, development and innovation (RD&I) investments towards achieving the technological objectives set out in the Strategy, drawn from the set of RD&I guidelines of interest for defense;

  • promoting actions that advance national and multinational cooperation in RD&I;

  • pushing excellence by leveraging the MoD as an RD&I catalyst and integrating talent from all sectors.

ETID also highlights the importance to find a balance between the innovative use of mature technologies and the advances associated with the development of emerging technologies. The innovative use of mature technologies involves a novel idea that combines different mature technologies for impact, greater than the sum of its parts. The ETID supports deep technologies that offer promising technological edge but require longer time to market and significant capital investments. It does not involve radically new or complex RD&I actions, nor does it involve large investments, and the results are obtained in the short or medium term. This is the type of innovation that has the greatest impact on different sectors in the civil sector (transport, communications, etc.). As for the development of emerging technologies, they require intensive efforts in RD&I, with investments that only a few countries or corporations are capable of deploying. The results come in the long or very long term and are helpful to maintain a competitive edge. However, fully understanding the possibilities of these still emerging technologies or the timescales, in which they will really take hold, is a major challenge.

Specific AI use cases are mentioned in the sections below. However, from a strategic planning perspective, it is important to note that the ETID acknowledges the fact that the scientific community and defense-relevant industry players express their frustration with the current system, “much more than in other countries,” as stated by the ETID strategy. This may be due to Spain’s bifurcated technology ecosystem, with commercial stakeholders focusing on the general application of AI and defense companies and defense-focused research institutes dealing with defense AI. This represents a self-limitation in the field of defense AI.

2 Developing Defense AI

Spain wants to advance national defense AI solutions and looks at the EU and NATO as important frameworks to develop these solutions in cooperation with partners. This section addresses each of the three lines of effort separately. Overall, Spain emphasizes that defense AI development must be rules-based as this reflects the country’s general stance on the importance of regulation.

2.1 Developing Defense AI at the National Level

Spain does not yet have a publicly available strategy or roadmap on defense digitalization, defense AI or a defense data policy. However, the ETID provides a roadmap, which aims to develop emerging and disruptive technologies, where AI plays an important role, to address seven challenges: defense applications with high technical exigencies, defense against asymmetric threats, leveraging the civilian technological boost, promotion of skilled labor capabilities, energy sustainability, digital transformation, and initial technological readiness.

The ETID addresses these specific challenges with a set of technological tools, among which AI serves as an important enabling tool. Table 1 includes those areas where AI is mentioned.

Table 1 Contribution of defense AI to specific technology development initiatives in Spain
Full size table

RD&I plays a pivotal role in developing technologies to address these seven challenges. In this regard, the ETID outlines what should be accomplished across eleven different areas of activities identified in the document. The following five make particular reference to the role of defense AI:

  • Precision Effects

The first area is arms and munition; concretely the improvement of munition performance from advanced guidance and control devices, as well as increasing their autonomy through the incorporation of sensor data processing technologies, artificial intelligence, communications links, etc., achieving operational advantages mainly related to improved combat effectiveness, reduced logistical burden, or reduced potential collateral damage.

  • Sensors and Electronic Systems

The second area is sensors and electronic systems. This is divided into three main blocs, where AI plays an important role. First, in radars, AI is expected to be used for identifying non-cooperative objectives and automated reconnaissance, as well as to develop new algorithms for radar processing. The second bloc is the processing of sensors data: concretely, the development of algorithms that analyze the data obtained by different sensors to automatically detect, recognize or identify the presence of entities of significance and interest to the Spanish Armed Forces (SAF), thereby reducing the analysis burden on human operators. The third area is electronic warfare.

  • Data Analytics for Cross-Cutting Technologies

The third area is typical technologies for bases, installations, platforms, and combatants. Concretely, AI-powered data intelligence is aimed at ensuring predictive maintenance of platforms, such as accurately and reliably predicting the remaining lifetime of each component or system. This is expected to be used for ground-based, naval, and aerial platforms, outer space systems, integrated systems, and to support maritime situational awareness.

  • CBRN Defense

The fourth area is Chemical, Biological, Radiological and Nuclear defense (CBRN defense), where AI would be useful to assess data provided by unmanned remote detection systems at high speed and for the identification of CBRN atmospheres based on AI-powered data.

  • Information and Communications Technology (ICT)

The fifth area is the inclusion of AI for ICT development, although this area is less precise regarding the potential of AI.

It is mainly the private sector that drives defense AI developments in Spain, but the private sector constitutes a patchwork of heterogenous actors. The biggest group is represented by a few large defense companies that represent the bulk of activities. These are usually led by Indra Systems (a Spanish information technology and defense systems company), Navantia (a leading maritime security company), and Tecnalia (a private R&D company). In addition, leading foreign defense companies like Thales or Airbus have offices and R&D centers in Spain.

Research institutions constitute the second largest group. Most of them come from the public sector, and some of them are part of the military ecosystem, like the Instituto Nacional de Técnica Aeroespacial Esteban Terradas. Others are a conglomerate of SMEs that have joined forces in clusters, such as the Andalusian CT Ingenieros Aeronauticos de Automocion e Industriales, but these clusters mostly focus on civilian, rather than defense engineering and thus only occasionally contribute to defense solutions. The smaller group consist of start-ups that mainly work for large companies. Some of them have their own portfolios, clients and investors, work individually, and provide services to international clients. One example is Devo, formerly Logtrust, a technology company that developed the first real-time big-data-in-motion software, which collects and analyses big data in real-time. The company became Spain’s first cybersecurity unicorn. Devo also provides AI solutions. It secured financial backing from venture capital firm Insight Venture Partners—an investor in tech giants such as X (formerly Twitter), Wix, Shopify, Trivago and many others—and has signed a USD9.5M contract with the U.S. Air Force (Devo 2020).

To grow the small number of start-ups and SMEs working on defense AI in Spain, large companies have taken action to entice more small companies to become active in this technology field. For example, Indra Systems launched the “FCAS Challenge” (Future Combat Air System), to boost the innovation capacity of the entire Spanish industrial ecosystem to the maximum by jointly developing emerging and disruptive technologies within the framework of FCAS. The FCAS Challenge aims to attract SMEs with cutting-edge technologies or highly innovative projects, with the objective of collaborating in the development, promotion, and maturation of their technological proposals. AI, data management in distributed clouds, technologies for low observability sensors, simulation and representation of information, optical technologies, or radio frequency (RF) and microwaves, are some of the technologies of greatest interest to the initiative (Indra 2023).

2.2 Developing Defense AI at the EU Level

The EU is an important reference point for the Spanish defense industry. The DDP, for example, stipulates the ambition to internationalize the Spanish defense industry within the framework of the EU (Arteaga 2022a, b). The main goal for Spain is to push national solutions abroad through multi-nationalization in areas where the Spanish industry is a leader. The main challenge still is to leverage EU projects to support those Spanish stakeholders that are not leaders in their areas but may benefit from these EU-wide projects to close existing industry capability and capacity gaps.

Strategies presented above show that the development of defense AI is part of the three lines of effort that Spain is contributing to at the EU level. In the research phase, it started with PADR (Preparatory Action on Defense Research 2017–2019) and now with Research Actions under the European Defense Fund (2021–2027). In the development phase, it started with European Defense Industrial Development Programme (EDIDP 2019–2020), and currently with Development Actions under the European Defense Fund (2021–2027). In the acquisition phase, the financial toolbox of contributions remains essential for a joint view on defense AI.

Spain has been an early supporter of the EDF launched in 2021 (Fiott 2023) and has since actively participated in several projects. Under the PADR, Spain led three defense research projects and 19 Spanish entities were involved in projects totaling €70M. Under the EDIDP, Spanish entities led eight projects and 64 of its entities were involved in EDIDP projects, that amounted for a total of €371M.

The naval domain has represented the largest share of the investments over the two EDF calls with an amount of €540M–€449M being invested in ground combat capabilities and €439M in space. The maritime sector is one of the key industries for the Spanish defense ecosystem, and where Spain is a leader worldwide. In particular, the Spanish ecosystem considers that the use of AI will become vitally important as an indispensable support to the decision-making process in the new way of operating at sea and from the sea. Finally, the operational environment of the near future requires fully connected naval forces, across branches and between allies. In this regard, Spain expects AI to play an important role in reinforcing ideas originally developed for network-centric warfare and to advance cross-domain cooperative engagement capabilities.

Although important, it is difficult to assess the specific defense AI development goals for current EDF projects. Based on this author’s analysis, the following seven projects seem particularly relevant:

  • Frugal and Robust AI for Defence Advanced Technology (FaRADAI)

FaRADAI focuses on frugal learning. The consortium includes the Fundaction Tecnalia Research & Innovation, Indra Sistemas, Thales Programas de Electronica y Comunicaciones and the Universidad Politecnica de Madrid.

  • Proactive automatic imagery intelligence powered by artificial intelligence exploiting European space assets (IntSen2)

IntSen2 develops AI for Imagery Intelligence (IMINT). The consortium is led by Tracasa Instrumental and includes Trabajos Catastrales and Zabala Innovation Consulting from Spain.

  • Knowledge Extraction, Machine Learning, and other AI approaches for secure, robust, frugal, resilient and explainable solutions in Defence Applications (KOIOS)

KOIOS seeks to improve defense AI, among other things, by defining metrics to measure frugality, robustness, and resilience. Led by CT Ingenieros Aeronauticos de Automocion e Industriales, the consortium also includes the Barcelona Supercomputing Centre, and NTT Data.

  • Convoy Operations with Manned-unManneDSystems (COMMANDS)

COMMANDS is to develop capabilities for cooperative manned-unmanned land systems. Under the leadership of Sener Aerospacial Sociedad Anonima, the consortium also includes Indra Sistemas and the Instituto Nacional de Tecnica Aerospacial Esteban Terradas and Santa Barbara Sistemas.

  • Digital Ship Structural Health Monitoring (dTHOR)

dTHOR aims to improve battle damage and structural integrity assessments of ships and includes the National Institute of Aerospace Technology, SAES, and Tecnicas Y Servicios de Ingenieria from Spain.

  • European Digital Naval Foundation (EDINAF)

EDINAF is about integration of a joint naval operational cloud in cloud-based approaches for Multi-Domain Operations. Led by Spain’s Navantia, the consortium also includes Aertec Solutions, Indra Sistemas, and the Universidade da Coruna from Spain.

  • European framework and proof-of-concept for intelligent automation of cyber defence incident management (EU-GUARDIAN)

EU-GUARDIAN shall create AI-based solutions to automate incident management and cyber defense. Indra Sistemas leads the consortium, which also includes the University of Murcia.

2.3 Developing Defense AI at the NATO Level

Spain’s commitments to its NATO allies have made it necessary to double the defense budget until 2029, starting with a 25.8% increase in 2023 (from €9.791bn in 2022 to €12.317bn in 2023). Of this, €5.241bn will be allocated to modernization programs (122A) and special modernization programs (122B) in 2023, €5.908bn in 2024, €5.576bn in 2025 and €5.766bn in 2026 (Fiott 2023). Unlike recently, the budget increase in investment should not only be aimed at providing military capabilities to the armed forces and orders to the defense industry (spend more), but also at adapting the national defense economy to the geostrategic rivalry with Russia and geopolitical competition with China (spend better), including AI.

Spain fully supports and contributes to NATO’s recent innovation initiatives, the Defence Innovation Accelerator for the North Atlantic Alliance (DIANA), and the NATO Innovation Fund (Ricart 2023). Spain is home to five test centers, which focus on AI alongside maritime security and defense, neurotechnology, 5G, quantum communications and energy. Spain expects both NATO defense innovations to stimulate the local AI-related defense innovation ecosystem and to advance interconnections with partners across the Alliance.

3 Organizing Defense AI

The governance of defense AI is managed by the Sub-Directorate General for Planning, Technology, and Innovation (SDG PLATIN) at the Directorate General for Armament and Materials (DGAM), which belongs to the Secretary of State for Defense at the Spanish MoD. The main function of SDG PLATIN is to propose and direct the plans and programs for research and development of weapons systems and equipment of interest for national defense, in coordination with the relevant national and international organizations. The SDG PLATIN is also responsible for:

  • Developing defense R&D policy;

  • Drawing up the Defense Technology and Innovation Strategy (ETID);

  • Coordinating and monitoring the defense R&D activities carried out by the different R&D centers of the MoD and establishing, coordinating, supervising, and evaluating the results achieved by the Department’s organizations executing the established R&D policy.

While defense AI is located in SDG PLATIN at the MoD, ETID, which provides guidance on the development of defense technology, is a sector-specific strategy derived from the overall Spanish Strategy on Science, Technology, and Innovation. This latter strategy is managed by the Ministry of Science and Innovation, which coordinates the National Plans of Scientific, Technical and Innovation Research (PEICTI) for the periods 2021–2023 and 2024–2027 that fund the technological goals of different ministries, including those of the MoD.

This means that there are two levels of inter- and intra-ministerial governance. First, ETID is an in instrument of inter-ministerial governance, focusing on defense AI and defense technologies. While the MoD is in lead, it acknowledges the need to promote and facilitate innovation and dual developments in coordination with other ministries, mainly the Ministry of Industry, Trade and Tourism (MINCOTUR), the Ministry of Science and Innovation (MICI) and the Ministry of Finance and Public Administrations (MINHAP).

In addition, the Secretariat of State for Digitalization and Artificial Intelligence also addresses AI. The Secretariat has existed since 2020 under the 1st Vice-Presidency and Ministry for Economic Affairs and Digital Transformation and has been transformed into the Ministry for Digital Transformation in late 2023. The National Strategy on Artificial Intelligence (Ministry of Economic Affairs and Digital Transformation 2020) emphasizes the need to leverage the opportunities and mitigate the risks presented by AI. This strategy does not include defense but earmarked €50M for investment in five flagship AI R&D projects in 2021 in agriculture and food; health; environment; energy; and employment, thereby covering 78 entities.

Second, intra-ministerial defense AI governance is organized around the SDG PLATIN. In addition, other Secretaries of State as well as other subordinate bodies such as the University Centers for Defense (CUD), the Research Centre for High Studies on National Defense (CESEDEN) and technical R&D centers affiliated to the MoD like the Esteban Terradas National Institute for Aerospace Technology (Instituto Nacional de Técnica Aeroespacial Esteban Terradas) play important roles. As RD&I involves the resources of many different departments, coordination via DGAM must involve many different stakeholders. Therefore, infrastructure-related activities involve coordination with the Directorate General for Infrastructure (DIGENIN), while ICT-related projects need to be synchronized with the Center for Information and Communications Systems and Technologies (CESTIC). Close coordination is not only needed to make best use of limited defense funds, but also in view of making sure that civilian RD&I efforts are not duplicated.

Moreover, Spain’s defense AI governance also needs to respect the country’s federal structure. Spain’s administrative-political system is composed of seventeen autonomous regions (“Comunidades Autónomas”) and two autonomous cities (Ceuta and Melilla in North Africa), with both exclusive competencies and competencies shared with the central government in Madrid. Defense is an exclusive competence of the central government. However, when it involves opportunities to perform defense AI R&D, there are several autonomous regions that harbor relevant institutions and research centers. For instance, Madrid hosts the EU Satellite Center. Valencia hosts a High Availability Military Base Camp with a long NATO record. Andalusia counts on a large network of defense companies, mostly in the maritime domain.

Finally, AI governance also includes complex discussions on ownership of sensitive data and involves non-defense entities engaged in dual-use activities. For example, the Government of Spain (Government of Spain 2023) announced the launch of the ADIA Lab, a research lab on AI, data science, machine learning, high-performance computing and quantum that would be based at the University of Granada, Andalusia. It receives funding from the Emirati sovereign wealth fund Abu Dhabi Investment Authority (ADIA). The project will last for four years, until 2027, and the goal is to support this research development to provide market solutions. There are no explicit statements on whether the knowledge transfer is bidirectional or goes to one specific direction. ADIA Lab is part of a bilateral agreement signed by the Presidents of Spain and the United Arab Emirates in February 2022. While there is no publicly available information on the lab’s concrete AI agenda and possible future use cases, the signing of the agreement was controversial with civil society organizations voicing concern over the risks of potentially sharing sensitive information with a non-EU and non-NATO country.

4 Funding Defense AI

Spain funds activities in support of defense AI via specific defense programs as well as programs that help to advance the country’s technological capabilities in general. In 2020, the new ETID set out the basic goals for promoting defense-related RD&I focusing on

  1. 1.

    maintaining participation in EDA R&T activities, especially in defining research priorities and agendas, and participating in cooperation projects when they effectively complement R&T activities carried out at the national level or in the EDF format;

  2. 2.

    increasing national participation in activities of the NATO Science and Technology Organization (NATO STO) following the main technology trajectories set out in the ETID, and improving the exploitation of their results, exploring and exploiting the possibilities offered by other NATO-promoted research and practical technological cooperation tools in areas other than STO activities, such as partnerships;

  3. 3.

    participating in the elaboration of EDF work programs by submitting proposals relevant for collaborative European defense R&D;

  4. 4.

    Supporting the EDF’s research window up to Technology Readiness Level (TRL) 6 and with a focus on domains that ensure sufficient technological capacity for the participation of Spain’s DTIB;

  5. 5.

    Supporting the EDF development window with the launch of priority technological developments (above TRL 6) of high cost and complexity, including supporting multilateral programs that offer sufficient opportunities for the Spanish DTIB to play an active role, and promoting standardization and interoperability of systems at European levels.

4.1 Funding Programs on Defense Solutions, Including AI

The MoD leads several funding programs that revolve around solutions for the defense ecosystem. Out of these, the Program for Cooperation in Scientific Research and Strategic Technologies Development (COINCIDENTE) aims to take advantage of civilian technologies developed within the scope of the National R&D Plan to incorporate innovative technological solutions of interest into the defense ecosystem.

COINCIDENTE projects must aim at developing a demonstrator with military functionality and involve significant technological innovation that meets a real or potential need of the MoD. The MoD’s co-funding level varies between 20% and 80% depending on general defense interest, the level of innovation of the respective project, its technological maturity, as well as the size of the company presenting the project.

The COINCIDENTE framework was created in 1985, but its calls did not include any reference to AI until 2018, when AI was requested as an essential element in proposals for projects implementing smart systems for military information analysis and solutions for other military problems. However, what was meant by “military problems” remained unclear. The call in 2020 included a specific requirement for detection technologies for active protection systems. The call in 2023 made a step forward and asked explicitly for solutions for synthetic data generation to train AI algorithms in defense applications.

Apart from the COINCIDENTE program, the MoD and other institutions have opened specific calls for concrete solutions that need short-term applications and responses on a non-annual basis. For instance, they launched Preliminary Market Consultations for the development of dual-use technologies, such as Earth satellite observation in 2023. The Centre for Technological Development and Innovation (CDTI), in collaboration with the Sub-Directorate General for Planning, Technology and Innovation of the DGAM, launched an initiative to promote the process of Innovative Public Procurement, in the form of pre-commercial procurement (PCP) in the field of Earth satellite observation. DGAM identified certain dual technology needs related to those that cannot be met through existing solutions on the market, and which could be addressed through a PCP process. One of the two proposals aims to integrate AI on board the satellites to reduce the necessary data transmission bandwidth.

4.2 Funding Technological Solutions, Including AI, with No Focus on Defense But Potential Alignment or Dual Use

Spanish institutions fund the development and implementation of AI. The CDTI funds several programs for international technological cooperation, including AI. CDTI organizes these international projects in two blocs. First, the Plan for International Calls, amounting to a total of 20 annual calls. Second, multilateral programs, that are made up of three initiatives:

  • Eureka is an intergovernmental undertaking that aims to support the development of products, services and processes and help these companies position themselves in third countries’ markets. To do so, Eureka establishes a partnership between a Spanish company and, at least, one firm or research center from a country that is part of the Eureka network. In addition to EU member states and associated countries, the Eureka Initiative also includes Ukraine, Argentina, Chile, Singapore, and South Africa.

  • IBEROEKA projects are open for partners from Latin American and Caribbean countries, plus Spain and Portugal.

  • Projects within the framework of the Partnership on Research and Innovation in the Mediterranean Area (PRIMA) are oriented towards Arab countries in North Africa, Turkey, and Southern European countries.

Out of these initiatives, AI tends to be included in the Annual Plan for International Calls and in then Eureka Initiative. However, it is difficult to assess if and to what extent projects funded under these initiatives generate spin-offs relevant for the Spanish defense ecosystem, as there is no breakdown on the specific applications of joint programs on AI.

5 Fielding and Operating Defense AI

5.1 Positive and Negative Goals

This section argues that the Spanish defense industry is guided by two main goals of the Spanish public administration: a positive goal that strives to seize the benefits of defense AI, and a negative goal that addresses the risks and challenges pertaining to developing and using defense AI. The Spanish government emphasizes both aspects as appropriate solutions for risk mitigation which can also create market opportunities for Spanish industry.

As for the positive goal, representative use cases include:

  • Electronic warfare solutions adapted to the current and future electromagnetic environment, inter alia with a focus on developing state-of-the-art electronic warfare systems in the non-communication and communication bands; electronic support (ESM) and electronic countermeasures (ECM), for which technological advances in antennas, components and RF modules will be applied; and advanced warning and intelligence algorithms adapted to the signals present in the electromagnetic (EM) environment.

  • Robotics with a focus on unmanned ground platforms for defense missions.

  • Industry 4.0 technologies to enhance the digital transformation of the MoD.

  • Automatic and intelligent analysis of large volumes of sensor data and technologies to enhance predictive maintenance of defense platforms.

  • Intelligent analysis of multiple sources of information for decision support.

  • Unmanned ground platforms for defense missions, unmanned surface, and underwater vehicles for defense missions as well as innovative applications of remotely piloted air systems (RPAS) for defense missions, and

  • Human Performance Modification with the use of exoskeletons for defense missions.

  • The use of AI in combination with sensors, signal intelligence, smart communications, and combat clouds as part of the FCAS Challenge launched by Indra Systems.

The negative goal covers the need for novel technological solutions to mitigate the risks of defense AI, due to the

  • lack of clarity in the decision logic of many AI algorithms, in particular deep learning algorithms;

  • difficulty in detecting fake multimedia content generated from real multimedia content;

  • lack of robustness or malicious introduction of training data that alters or biases the learning process;

  • data security and privacy throughout the process;

  • unpredictable behavior of algorithms in novel situations for which they have not been prepared;

  • possible cascading failures when incorporating multiple AI-based software modules in a complex system, such as those used in defense;

  • need for large volumes of data to train algorithms;

  • complexity of incorporating ethical criteria into decision-making processes.

Moreover, the ETID 2020 addresses the applications of biometric technologies, and the need to test the applications for speech and text analysis.

5.2 Technology Readiness Levels

For fielding and operating defense AI, there are two levels of action: early-stage technology development, and highly advanced technology maturity. For the first area of action, the main goal is to foster research in emerging technologies for TRL 1–3 (early stages) and achieve a growing level of complementarity with specific instruments managed by the public administration through calls for projects. As for high-technology maturity, the goal is to foster technology demonstrators (TRL 4–6) and prototypes (TRL 7–8) and leverage civilian developments to be incorporated into the military field with dual-use approaches.

However, what seems straightforward in theory looks more challenging in practice as several hurdles prevent the members of the Spanish defense ecosystem from fully exploiting the opportunities of defense AI (Díaz 2020). These hurdles include:

  • Access to Information

The MoD prepares, in a cyclical process every six years, its defense planning cycle, which results in two documents, the Long-Term Doctrine Objective 2035 (Ministry of Defense 2019) and the Military Capabilities Objective. However, due to classification both documents are not publicly available. Therefore, members of the Spanish DTIB do not have access to the armed forces’ short-, medium-, or long-term needs. Consequently, investments and defense development priorities may not be consistent with the needs of the Armed Forces.

  • Access to Requirements

The detailed set of requirements formulated in operational, functional, technical, logistical, and physical terms to be met by these systems is not available. DTIB developments may not meet these requirements and, consequently, may not be of use to the SAF. Access to data can be included in this group when it is relevant to the design of solutions, such as in the case of systems that include AI.

  • Access to Evidence

Information about Spanish defense projects are not always available to all members of the national ecosystem. Information asymmetry, however, is a problem as it can stifle competition and prevent the development of defense solutions that meet the requirements of the armed forces. Timely access to information would also help speed up development cycles. But demands for comprehensive information access collide with information classification.

  • Access to Operational Validation Environments

Quality assessment criteria are needed when developing and deploying new defense solutions. In this regard test and experimentation centers play an increasingly important role to verify what has been developed. The number of these test and experimentation centers is growing in Spain, but they are not yet well spread across the country.

5.3 A Bet on Transparency in Operationalizing Defense Technologies

Although the operationalization of defense technologies projects is evident, there is still a lack of comprehensive awareness of investments undertaken at the national level. To this end, the public sector indicates that it has collaborated with private firms to establish quantitative indicators to assess the level of investments devoted to RD&I projects meeting defense goals, the number of projects and mobilized economic volume in dual-use projects, and the number of projects with private non-governmental funding. In addition, indicators also set out to assess quantitively the number of Spanish entities participating in the European DTIB, the percentage of economic returns generated by Spain’s participation in EDF projects, the number of projects that applied to the EDF with Spanish participation, the results (granted, non-granted), and the investments in other multinational initiatives on technology development and investments returns. At the time of writing this chapter, the respective figures are still being collected.

6 Training for Defense AI

There is not much publicly available information on training for defense AI. However, this section aims to divide this topic in two main parts. First, how defense AI systems are trained. Second, how the human capital is trained to design, develop, implement, deploy, and use defense AI.

6.1 Training the Systems

Three lines of effort are worth mentioning to train defense AI systems:

  • Training the Systems for Specific Projects

First, the public sector develops its own projects. This is the case of the Esteban Terradas National Institute for Aerospace Technology (Instituto Nacional de Técnica Aeroespacial Esteban Terradas, INTA), which conducts self-funded R&D projects in different technology areas (Marín 2019). IDATECT is the project that makes use of AI to improve the competences in the areas of Experimental Aerodynamics and Theoretical and Computational Aerodynamics to contribute to meeting the following technological challenges: new configurations with improved aerodynamic efficiency leading to lower fuel consumption and thus pollution reduction; innovative concepts for aerodynamic efficiency improvement, including active flow control devices, high-lift devices; multilevel and multidisciplinary modelling and simulation methods; robust and reliable design strategies, in order to manage the uncertainties of the models and their input data; and big data and artificial intelligence, including techniques that use data extracted from various sources to enable efficient and fast decision-making.

  • Creating Specific Centers for AI Training

Training the AI systems for defense is not only a matter of funding specific projects. INTA’s 2024 Work Plan (INTA 2023) foresees establishing the new Technological Center for Development and Experimentation (CETEDEX), located in Andalusia (Jaen). CETEDEX will also house three new centers aimed at developing anti-drone technologies, general vehicle and connected vehicle technologies, and AI. In 2024, ongoing work on setting up the main campus will continue, while the Advanced Proving Ground Project and initial technology projects will be launched.

  • Training the Systems to Improve Intra-Institutional Performance

INTA is working to develop a project aimed at digitalising the Institute’s activity. This project is structured into four major blocks with one block focusing on establishing the AI Supercomputing Center to accommodate the needs for modelling and simulation, using the most modern AI techniques on a single platform. The center will also develop a cloud system in both a private and hybrid cloud, which will provide the necessary flexibility and capacity for future INTA digital transformation projects.

6.2 Training the People

There is limited information on specific trainings for defense AI. However, private defense companies offer hands-on training on the usage of AI, inter alia, with a focus on defense AI training related to maritime security, aerospace, and the linkage between cybersecurity and AI. In parallel, the public sector also seeks to reinforce trainings in practical terms. Already in the ETID 2020, the public administration suggested the creation of challenge programs and hackathons.

Integrating AI into defense solutions will require more technical coordination since AI—as a cross-cutting technology—will depend on the parallel modification of other technologies and specifications to generate added value for the SAF. This increases the level of complexity compared to traditional defense solutions. There is—in particular, against the systemic hurdles mentioned in the preceding section—a risk that unsynchronized development priorities will inevitably lead to imbalances in the technological levels of sophistication. The lack of publicly available information on training goals and activities undertaken by the MoD, the military services, and the defense industry underlines the pressing need to focus more attention on developing the skill sets needed to successfully design, develop, implement, deploy, use, and assess the results generated with defense AI in Spain.

7 Conclusion

Spain is slowly waking up to the realities of defense AI. AI has been considered an important issue across national strategies since the mid 2010s. However, it was not until the 2020s that AI became a centerpiece of strategy and was translated into roadmaps, tailored measures, and specific applications.

Although the governance system is adequately organized and coordinated across many different stakeholders, more needs to be done to advance defense AI in Spain. For example, funding needs to be increased, a broader portfolio of projects should be pursued, and information relevant for long-term force development in Spain needs to be made accessible to the Spanish defense industry. Furthermore, public, and private stakeholders need to engage more closely on setting out the goals that the use of defense AI is expected to accomplish, as well as the results (and the failures) ongoing defense AI development projects generate.

The defense market of defense AI is led by a few large companies that are also leading an important number of defense development projects at the EU level. Still, a larger focus on SMEs should developed to ensure that defense AI can tap into the capacities of a broad set of companies to avoid critical dependency on a few players.

The fielding and operationalization of defense AI benefits from taking place through different instruments, although further activities are recommendable due to the potential opportunities that AI brings into the defense ecosystem. This proposal would help increase the defense industrial footprint in Spain, create cross-company ecosystems, and increase the participation of Spanish companies in European projects. Finally, there is limited information on the training of systems and of human capital. A greater inclusion of curricula on defense AI should be fostered in university programs.

While it is not possible to define the long-term pathway for defense AI in Spain, the current trajectory shows that the ecosystem, the number of activities and the interest are growing, suggesting a gradual uptick of defense AI activities in the country. However, the main challenge remains sustained investments in the technologies and skillsets needed to develop defense AI.